Topic: A4 Dominator - Pre-Order Group Buy - 280mh, roughly 1000w, $1800 + shipping - page 45. (Read 122571 times)

This rig is TOO NEW to be able to give any kind of real performance metrics.  Those of us with these miners bought them as a pre-order before the miners were even built and we have only had them for a short while.  

Since it is so new there are some challenges that need to be figured out with the software it appears.  Reliability is yet to be determined.  My personal recommendation at this time is for people to wait until we sort out these problems.  Right now the miners are crashing, dropping cards, and not quite hashing at the advertised rates.  So if it were my money and I was looking to buy right now(knowing what I know right now) I would be saying wait and see if we can fix these issues that we are having.  Otherwise you are going to get yourself a full time job babysitting miners trying to keep them running.

I have seen periods where the miner runs fine for a while.  I think my longest run has been like 18-19hrs.  But once it drops a card, or starts the crash reboot cycle then things get ugly.  So it is hit or miss at the moment.  

45 day warranty?!

...and their website is offline?!

that's pretty sketchy!

what are people's experiences regarding mining with this machine?

it's been suggested to try 825mv -- might give that a go.

That being said, over 9 hours now on Prohash without an issue here using 1.0.0T

EDIT: guess I spoke too soon lol -- 1 card went down, not minutes after this post and the miner restarted itself. -- Now trying 820mv

Ok, I have lost count, back to back crash and reboots all day long now.  The settings in 1.0.2 is 815mv.  I have not changed it.  At this rate I will have to go back to the stock image.  The 1.0.2 is more unstable than the first build from what I am seeing.

If you have the ability to run 1.0.0T I suggest switching to that, after switching back to that, mine is at least somewhat stable.

What voltage are you running. And what is your stale shares percent

Well I don't know what transpired, but I just looked over at the A4 screen and 3 boards were offline/missing.  So I don't know if they just stopped hashing, or if the miner detected a board going offline and restarted the miner and 3 cards just didn't come up.  I don't know.  I was busy with some other things.  So I just stopped the miner, restarted it, and 2 more came back online, stopped and restarted again and all 4 boards are back online.  

So clearly 1.0.2 does not fix anything.  I will leave it on litecoinpool until we an get something that is more stable.  No way this thing is going to stay alive for very long on a multipool.   So at this point all I can do is try to keep an eye on this thing and try to keep it running.  Not a good situation at all.  It is not ready for large scale mining farm deployment unless you have someone watching these things all the time.  

Anyone thinking about getting the A4's should hold off until we see if we can get these issues sorted out.  Unless of course you can babysit the A4's full time.

UPDATE: After starting the miner and getting all the boards online one of them died again and miner has restarted again on it's own.  This time bringing all 4 boards online and hashing normally at this point.  I have noticed a pattern when starting the miner that it appears to do this often.  So watch your miners closely when you first start or restart them.  It will take about 15-20 minutes for the failing card to catch the attention of the software and cause a reboot.  

UPDATE2: Yet another restart on the miner. 

Well the A4 managed to stay up overnight.  I will just let it run and see how long it will go before it restarts.  I am still on Litecoinpool.  If it will run at least 24 hours then I may try my hand again on Prohashing to see how it reacts to getting it's feathers ruffled with coin switching... When I tried with the stock software things did not go well. 

My cables for the PSU didn't arrive yesterday so I won't get to play with the second A4 until Tuesday.  Was really hoping to get the second one online this weekend. 

I have switched back to older build.

Apparently I was wrong about 1.0.2 being slower, I switched back to 1.0.0T and it was slower 6 minutes in. I tried to take a 2nd picture of 1.0.2 at about 25 minutes but it crashed right in front of me while I was attempting to screen cap. EDIT: I found on another site that someone said 1.0.2 restarts around 22 minutes -- this corroborates what I experienced.

I believe I have a Batch 1 based upon posts I found else where and the fact i have the 1.0.0T version.



1.0.2 Screen capture from previous post



1.0.2 @ ~6 Minutes -- Hashing around 272MH


1.0.0T @ ~6 Minutes -- Hashing around 267MH


At this point, I'll trade speed for (more) stability between the 2 unstable versions I have.

I have 3 running miners. Each of them drops 1 or 2 cards at least once per day.
Fortunately, I have a switched PDU which allows me to power cycle remotely.
I need to "automate" that somehow, because I don't catch the drops when they happen, only if I go look at the miner.
Im running the G5-1102 version.

Well 1.0.2 is less stable for me than the version I was running previously.

When the miner drops a card, it absolutely will not restart on 16384 starting difficulty, if I change it to 8, it starts right up.

I'm still doing some testing but I think I'm going back to the previous version unless I am able to find some stability.

At least on the previous version it would at least run like 2 days before a card would drop.

 There were 2 folks that wrote A2 software upgrades (This is NOT Firmware, it's SOFTware).

 I'm sure one or both will be looking at A4 software "fixes" as soon as they get an A4.

I'm seeing the same thing with my Batch 2 miners. These are the miners with the controller software with V1.0.1 beneath the LTC LIMITED--T SERIES and can run the V1.0.2 image Chloe posted. The Batch 1 miners I have run rock solid with the 1.0.0 firmware (just says LTC LIMITED--T SERIES with no version number beneath). The hardware is different in some way. A B2 image will not run on a B1 miner. Hashrate = 0

If you're not sure what you have or mixed them up like I did, you can find the manufacturing date on the tamper proof sticker which is covering one of the rear case screws. Batch 1 machines were made before 15th Oct.

At the moment I can't keep a batch 2 machine with the new V 1.0.2 firmware running for longer than about 30 minutes before it restarts itself. I'm glad to hear that Inno will open source the firmware. It would be good to see Gen Tarkin or Emdje work on improving it.

I'd happily sponsor it with a few Satoshis.

No doubt in my mind that v1.0.2 is a fraction slower.  still have blades dropping though. Not sure what this version was suppose to fix.
Fortunately, im running them on a switched pdu, so I can power cycle them from anywhere.

Confirmed here too, 1.0.2 does not fix the problem for me, if anything my experience is worse on this build than the build I was running. 1 card went down after only a few hours.

I noticed that the miner was most stable for me when I set starting difficulty at 8... 1.0.2 was built with 8192 as diff setting but still showing 8 as default. After I stopped the miner in an attempt to get the missing card to come back online, it would not restart.

I checked all my settings and changed the starting difficulty back to 8 and it came up.

Don't know if this will have an effect for anyone else or if it was just dumb luck.

On a side note, I feel like 1.0.2 hashes just slightly lower than the version I was previously using (I think it was 1.0.1)

how long has there site been down?

 http://www.innosilicon.com/

Thats what worries me most either dealing with poor coding skills or inherent hardware design flaw.

Running new build, it came configured with 815mv

So it appears the v1.0.2 rebooted again on me.  Not sure what precipitated it as I was out running errands when it happened.  So it does not appear the newer firmware fixes problems with the need to restart.  At least if it is due to a hung card it appears to be trying to keep the miner alive at least.  But stable? Nope not from what I am seeing.

I agree